The Illusion of AI Safety
Chatbots answer 30 prompts about self-harm, eating disorders, and substance use. Researchers check replies for harmful content and invitations to continue the conversation.
Slingshot AI reran the prompts with four models and evaluated 24,000 responses across the initial and follow-up turns.
- Released
- October 2025
- Kind
- Wellbeing and sycophancy
- Level
- University
- Scored
- Graded by people
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Propensity
- Behavior
- Misaligned with values
- Language
- English
- Interaction
- Multi-turn
- Items from
- Written by experts
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Content safety
- Strongest model tested
- GPT-5 #71 on the capability index
- Last model added
- August 2025
- Scores refreshed
- November 21st, 2026
GPT-4o leads with 43%
Best score per model from the benchmark's paper, read on November 21st, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best.
- 1 GPT-4o Mar 2025 43%
- 2 GPT-5 53%
Source: The The Illusion of AI Safety paper
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-4o Mar 2025 | 43% | Mar 2025 | Official leaderboard | |
| 2 | GPT-5 | 53% | Aug 2025 | Official leaderboard |
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official The Illusion of AI Safety leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6%
- Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts.
Gemini 4 Carbon 0%
- Covert misbehavior (Thinking Machines) In use Agentic tasks that check whether a model misbehaves and conceals it. Scored by the rate of covert misbehavior.
Inkling 2 Mini 8.30%
- Destructive actions (Thinking Machines) In use Agentic tasks that check whether a model takes destructive actions it was not authorized to take. Scored by the rate of destructive actions.
Inkling 2 Small 4.70%
- Misbehavior (Thinking Machines) In use Agentic tasks that check whether a model fabricates results, ignores instructions or hides what it did when a task is hard. Scored by the rate of misbehavior.
Inkling 2 Mini 9.80%
Data audit
7 of 8 checks pass, 1 warning. Look first at the Newest score check. The newest came 15 months ago, in August 2025, for GPT-5. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 1 pass 3 notes
- Pass Independent runs Scores from the benchmark’s paper (2 entries).
- Note Models scored Only 2 models we track.
- Warning Newest score The newest came 15 months ago, in August 2025, for GPT-5.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
- Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
- Pass Importer We read the board on counterharm.com: 2 rows for models we track at the last read.
- Pass Last read November 21st, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.